Published on Web 10/10/2003
Ultrafast Excited-State Dynamics of Adenine and
Monomethylated Adenines in Solution: Implications for the
Nonradiative Decay Mechanism
Boiko Cohen, Patrick M. Hare, and Bern Kohler*
Contribution from the Department of Chemistry, The Ohio State UniVersity,
100 W. 18th AVenue, Columbus, Ohio 43210
Received April 14, 2003; E-mail: [email protected]
Abstract: The DNA base adenine and four monomethylated adenines were studied in solution at room
temperature by femtosecond pump-probe spectroscopy. Transient absorption at visible probe wavelengths
was used to directly observe relaxation of the lowest excited singlet state (S1 state) populated by a UV
pump pulse. In H2O, transient absorption signals from adenine decay biexponentially with lifetimes of 0.18
( 0.03 ps and 8.8 ( 1.2 ps. In contrast, signals from monomethylated adenines decay monoexponentially.
The S1 lifetimes of 1-, 3-, and 9-methyladenine are similar to one another and are all below 300 fs, while
7-methyladenine has a significantly longer lifetime (τ ) 4.23 ( 0.13 ps). On this basis, the biexponential
signal of adenine is assigned to an equilibrium mixture of the 7H- and 9H-amino tautomers. Excited-state
absorption (ESA) by 9-methyladenine is 50% stronger than by 7-methyladenine. Assuming that ESA by
the corresponding tautomers of adenine is unchanged, we estimate the population of 7H-adenine in H2O
at room temperature to be 22 ( 4% (estimated standard deviation). To understand how the environment
affects nonradiative decay, we performed the first solvent-dependent study of nucleobase dynamics on
the ultrafast time scale. In acetonitrile, both lowest energy tautomers of adenine are present in roughly
similar proportions as in water. The lifetimes of the 9-substituted adenines depend somewhat more sensitively
on the solvent than those of the 7-substituted adenines. Transient signals for adenine in H2O and D2O are
identical. These solvent effects strongly suggest that excited-state tautomerization is not an important
nonradiative decay pathway. Instead, the data are most consistent with electronic energy relaxation due to
state crossings between the optically prepared 1ππ* state and one or more 1nπ* states and the electronic
ground state. The pattern of lifetimes measured for the monomethylated adenines suggests a special role
for the 1nπ* state associated with the N7 electron lone pair.
Introduction
absorption1,2
Femtosecond transient
and fluorescence upconversion3-6 measurements have recently established that the DNA
and RNA bases have subpicosecond fluorescence lifetimes in
aqueous solution at room temperature. Ultrafast relaxation of
electronic energy provides the building blocks of life with a
high degree of photostability and reduces the likelihood of
photochemical damage. In particular, relaxation to the electronic
ground state on the femtosecond time scale greatly reduces the
yield of reactive triplet molecules. Considerable effort is
currently devoted to understanding nonradiative decay by the
nucleobases,7-12 but consensus on the mechanism has yet to
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
Pecourt, J.-M. L.; Peon, J.; Kohler, B. J. Am. Chem. Soc. 2000, 122, 9348.
Pecourt, J.-M. L.; Peon, J.; Kohler, B. J. Am. Chem. Soc. 2001, 123, 10370.
Peon, J.; Zewail, A. H. Chem. Phys. Lett. 2001, 348, 255.
Gustavsson, T.; Sharonov, A.; Onidas, D.; Markovitsi, D. Chem. Phys. Lett.
2002, 356, 49.
Gustavsson, T.; Sharonov, A.; Markovitsi, D. Chem. Phys. Lett. 2002, 351,
195.
Onidas, D.; Markovitsi, D.; Marguet, S.; Sharonov, A.; Gustavsson, T. J.
Phys. Chem. B 2002, 106, 11367.
Broo, A. J. Phys. Chem. A 1998, 102, 526.
Nir, E.; Kleinermanns, K.; Grace, L.; de Vries, M. S. J. Phys. Chem. A
2001, 105, 5106.
Rachofsky, E. L.; Ross, J. B. A.; Krauss, M.; Osman, R. J. Phys. Chem. A
2001, 105, 190.
13594
9
J. AM. CHEM. SOC. 2003, 125, 13594-13601
be achieved. Only a small number of nucleobases have been
studied on the ultrafast time scale, and additional experiments
are sorely needed to evaluate proposed mechanisms.
Here we present a systematic study of adenine and adenine
derivatives in solution at room temperature. The femtosecond
transient absorption technique was used to study adenine and
all four monomethylated adenines in which a methyl group is
attached to a purine ring nitrogen (see Chart 1). Because the
methyl group is not labile, each of these compounds provides
insight into the photophysics of an adenine tautomer with
hydrogen at the same position. There is longstanding interest
in the photophysics of different adenine tautomers since it has
been argued that most of the fluorescence is emitted from a
minor tautomer.13 As part of our findings, we report the first
lifetime measurements for the 7H- and 9H-amino tautomers of
adenine. Even though 7,9 tautomerism does not occur in DNA,
adenine is an important model system for understanding
(10) Mennucci, B.; Toniolo, A.; Tomasi, J. J. Phys. Chem. A 2001, 105, 4749.
(11) Sobolewski, A. L.; Domcke, W.; Dedonder-Lardeux, C.; Jouvet, C. Phys.
Chem. Chem. Phys. 2002, 4, 1093.
(12) Ismail, N.; Blancafort, L.; Olivucci, M.; Kohler, B.; Robb, M. A. J. Am.
Chem. Soc. 2002, 124, 6818.
(13) Wilson, R. W.; Callis, P. R. Photochem. Photobiol. 1980, 31, 323.
10.1021/ja035628z CCC: $25.00 © 2003 American Chemical Society
Ultrafast Excited-State Dynamics of Adenines
ARTICLES
Chart 1. Amino and Imino Tautomers of Adenine
tautomerism. Rare tautomers may be responsible for substitution
mutagenesis.14 Additionally, slower rates of electronic energy
relaxation by rare tautomers could make them more susceptible
to photochemical damage. We previously pointed out the
approximate correlation between excited-state lifetimes and the
propensity for nucleobase photochemistry.2
An outstanding issue in nucleobase photophysics is the role
played by the solvent. This is particularly important given a
recent report of ultrafast lifetimes for isolated bases.15 We
therefore undertook the first study on the influence of the solvent
on nucleobase excited-state dynamics. Specifically, we characterized S1 dynamics in H2O, D2O, and in the aprotic solvent,
acetonitrile. Thus, the twin aims of this report are to comment
on intramolecular (tautomerism) and intermolecular (solvent
effects) factors that influence nonradiative decay in adenine.
The questions to be addressed include: (1) What are the
consequences of tautomerism for the excited-state lifetimes? (2)
What is the ground-state distribution of tautomers at room
temperature? (3) Is interconversion from one tautomeric form
to another on the excited-state potential energy surface the
mechanism behind nonradiative decay? (4) What do the
observed solvent effects imply about the mechanism of excitedstate relaxation?
Experimental Section
Transient absorption signals were recorded by the pump-probe
technique using our amplified titanium sapphire laser system, as
described previously.2 Briefly, pump pulses with a center wavelength
of ∼263 nm were obtained from the third harmonic laser output. Probe
pulses were derived from a white light continuum generated in a 1-cm
path length cell filled with water. Modifications to the stretcher and
compressor of our chirped pulse amplifier reduced the pulse duration
of the fundamental output to ∼90 fs, as measured by second-order
autocorrelation in a KDP crystal. The difference-frequency mixing
signal between the third harmonic pump pulse and the fundamental in
the probe arm measured at the sample position had a fwhm of 270 fs.
From this measurement, the fwhm of the instrument response function
is ∼190 fs, assuming Gaussian pulse shapes. The induced absorbance
change (∆A, equivalent to ∆OD) was recorded versus time delay
between pump and probe pulses. ∆A was calculated as -log(I/I0), where
I is the probe signal measured by a photodiode after the sample in the
presence of the pump pulse, and I0 is the same signal with the pump
blocked.
Aqueous solutions of the various adenines were studied primarily
in pH 6.8 buffer. The buffer was prepared by dissolving 1.77 g of Na2(14) Watson, J. D.; Crick, F. H. C. Nature (London) 1953, 171, 964.
(15) Kang, H.; Lee, K. T.; Jung, B.; Ko, Y. J.; Kim, S. K. J. Am. Chem. Soc.
2002, 124, 12958.
Figure 1. Normalized steady-state absorption spectra at pH 6.8 for adenine
(2), 1-methyladenine (0), 3-methyladenine (b), 7-methyladenine (×), and
9-methyladenine (O). The inset shows the extinction coefficients (units of
M-1 cm-1) of 7-methyladenine (×) and 9-methyladenine (O) at pH 6.8 as
a function of wavelength.
HPO4 and 1.70 g of KH2PO4 in 500 mL of ultrapurified water.
7-Methyladenine (7MA) was obtained from the National Cancer
Institute Chemical Carcinogen Reference Standard Repository. (The
standard ring numbering of adenine is shown in Chart 1.) 1-Methyladenine (1MA) and 9-methyladenine (9MA) were purchased from Acros
Organics, while adenine and 3-methyladenine (3MA) were from SigmaAldrich. All compounds were used as received. UV/vis spectra were
recorded frequently during experiments to assess photodegradation, and
solutions were changed as necessary.
Solute concentrations were generally adjusted to give an absorbance
between 1 and 2 for a 1-mm optical path length at our pump wavelength
of 263 nm. Adenine and the monomethylated derivatives studied here
are, however, significantly less soluble in acetonitrile than in water,
and this led to solutions of lower absorbance. Acetonitrile solutions
were prepared by adding excess solute to a quantity of solvent, stirring
overnight, and finally filtering out the undissolved solid. The optical
densities of these saturated solutions (measured at the pump wavelength
of 263 nm for a 1-mm optical path length) were 0.24, 0.18, and 0.49
for adenine, 7MA, and 9MA, respectively. Saturated solutions of 1MA
and 3MA in acetonitrile had optical densities <0.1 and were too dilute
to permit pump-probe experiments.
Data Analysis. Signals were modeled as sums of exponentials
analytically convoluted with a Gaussian that represents the instrument
response function of our ultrafast spectrometer. Transients at several
probe wavelengths were fit simultaneously with the time constants of
the exponentials as global fitting parameters, as described previously.2
The fwhm of the Gaussian function was an adjustable fitting parameter.
This accounts for modest changes in time resolution that occur over
periods of weeks and months due to changes in laser operation.
Although the fwhm was adjustable, it changed little about its average
value. In no case did optimization of this parameter produce a value
that was less than 190 fs, the estimate of our best time resolution from
the difference frequency-mixing signal discussed above. Parameter
uncertainties were obtained from the formal covariance matrix after χ2
minimization and are twice the standard deviation (2σ), unless stated
otherwise.
Results
Adenine in H2O. Steady-state absorption spectra in H2O are
shown in Figure 1 for adenine and the derivatives used in this
study. All compounds show strong ππ* absorption at the pump
wavelength of 263 nm. The spectra are similar except for 3MA
and 7MA, which exhibit significant red-shifts. Transient absorption at 570 nm for an aqueous solution of adenine is shown in
Figure 2. The signals exhibit complex decays, which are
identical at long delay times with signals recorded on a solvent
J. AM. CHEM. SOC.
9
VOL. 125, NO. 44, 2003 13595
ARTICLES
Cohen et al.
Figure 2. Transient absorption at 570 nm by adenine in aqueous solution
(circles connected by a solid line to guide the eye). The dashed curve is
the transient absorption signal at the same probe wavelength from a blank
containing just water and the pH buffer. It has been scaled to agree with
the signal from adenine at long times. The figure inset shows the data at
short times. The pump wavelength for this and all subsequent figures is
263 nm.
blank containing just the pH buffer (dashed curve in Figure 2).
These long-time signals seen in our transient absorption
experiments arise from absorption by solvated electrons produced by the two-photon ionization of water.2 The very small
cross sections for excited-state absorption (ESA) by the nucleobases make it impossible to eliminate this signal, even at very
low pump powers. As reported previously,2 ESA by the purine
bases is stronger than by the pyrimidine bases. As a result, the
signal contribution from water is smaller for adenine than for
cytosine (compare Figure 2 with Figure 1 in ref 16). For the
adenines studied here, the water signal typically accounts for
<20% of the maximum signal strength. On the time interval
from ∼1-100 ps, the water signal is nearly constant due to the
slow time scale for geminate recombination by solvated
electrons.17
This signal component can be removed by subtracting the
signal measured on a solvent blank, after it has been scaled to
agree with the nucleobase signal at long times. We verified that
the long-time signal varies strictly quadratically with pump
power, consistent with two-photon ionization. In this study, fits
to the data used to determine lifetimes were always made to
solvent-subtracted signals. This procedure results in somewhat
more accurate time constants, especially when τ is comparable
to the rise time of the solvated electron signal (∼300 fs; see
Figure 2 inset). It also makes it easier to discern low-amplitude
decays that occur on slower time scales, as seen in the solventsubtracted data in Figure 3. In this and later figures, solventsubtracted data is indicated by ∆As on the vertical axis. The
transients in Figure 3 show relaxation at three probe wavelengths
across the ESA band of adenine. The maximum signal was
observed near 600 nm in good agreement with the transient
spectrum previously reported for cytidine.2 Global fitting of the
three transients in Figure 3 to a model function of two
exponential decays gave time constants of 0.18 ( 0.03 and 8.8
( 1.2 ps. Fit parameters are summarized in Table 1.
Monomethylated Adenines in H2O. Like adenine, the
monomethylated adenines all exhibit a weak and broad transient
(16) Malone, R. J.; Miller, A. M.; Kohler, B. Photochem. Photobiol. 2003, 77,
158.
(17) Crowell, R. A.; Bartels, D. M. J. Phys. Chem. 1996, 100, 17940.
13596 J. AM. CHEM. SOC.
9
VOL. 125, NO. 44, 2003
Figure 3. Transient absorption of adenine in pH 6.8 buffered aqueous
solution at 570 nm (O), 600 nm (0), and 650 nm (4). Signals have been
vertically offset for clarity. Solid lines are from a global fit to a biexponential
decay. The solvent-only signal has been subtracted as described in the text.
Table 1. Adenine Global Fit Parametersa
adenine in H2O
adenine in CH3CN
τ1/ps
τ2/ps
A1
A2
0.18 ( 0.03
0.44 ( 0.07
8.8 ( 1.2
11 ( 5
0.86
0.90
0.14
0.10
a Transients were fit using the function ∆A ) A exp(-t/τ ) +
1
1
A2 exp(-t/τ2). Time constants, τ1 and τ2, were global-fitting parameters,
but the amplitudes, Ai, were not. The Ai values in the table are for a probe
wavelength of 570 nm, and they have been scaled such that ∑i|Ai| ) 1.
Figure 4. Transient absorption at 570 nm of optically matched aqueous
solutions of 9-methyladenine (0) and 7-methyladenine (O) collected under
identical experimental conditions. Solid lines are from a nonlinear leastsquares fit to a monoexponential decay. The solvent-only signal has been
subtracted as described in the text.
absorption band centered near 600 nm. Transient absorption
signals for 7MA and 9MA at a probe wavelength of 570 nm
were recorded in back-to-back experiments at pH 6.8 (Figure
4). For the results in Figure 4, the 7MA and 9MA solutions
had concentrations of 1.71 × 10-3 M and 1.53 × 10-3 M,
respectively, resulting in identical absorbance (i.e., optical
density) of 1.6 at the pump wavelength of 263 nm. At constantpump laser intensity, equal absorbance solutions have identical
excited-state populations initially, allowing relative ESA cross
sections to be measured (see the Discussion section). We note
that the magnitude of the long-time signal (prior to subtraction
of the solvent-only signal) was the same for both, as expected
for equal absorbance (“optically matched”) solutions.
Signals from 1MA and 3MA are shown in Figure 5. The pH
for 1MA was adjusted to pH 9, while the pH of the 3MA
Ultrafast Excited-State Dynamics of Adenines
ARTICLES
Figure 5. Transient absorption of 1-methyladenine at pH 6.8 (O), and
3-methyladenine at pH 8.2 (0) at 570 nm. Solid lines are from a nonlinear
least-squares fit to a monoexponential function. The signal from water has
been subtracted as described in the text.
Figure 6. Normalized transient absorption signals for adenine (4),
7-methyladenine (O), and 9-methyladenine (0) in acetonitrile at 570 nm.
Global, nonlinear least-squares fits are shown by the solid curves.
Table 2. S1 Lifetimes (ps) of Monomethylated Adenines at Room
Temperature
solvent
1MA
(570a)
3MA
(570)
7MA
(570, 600, 630)
9MA
(570, 600, 630)
H2Ob
CH3CN
0.26 ( 0.03
c
0.18 ( 0.06
c
4.23 ( 0.13
3.3 ( 0.3
0.22 ( 0.02
0.35 ( 0.02
a Probe wavelengths (nm) of transients used in the fits are shown in
parentheses after each compound. Multiple wavelengths indicate a global
fit was performed. b The pH was 9.0 for 1MA, 8.2 for 3MA, and 6.8 for
the remaining compounds. c Insufficient solubility.
solution was adjusted to 8.2. These higher pH values were used
to ensure that both compounds were present in their neutral,
unprotonated forms on account of their higher basic pKa’s of
∼7 for 1MA and ∼6 for 3MA.18 Because of the very weak
signals from these compounds, we recorded only a single
transient at 570 nm near the maximum of the ESA band. The
lifetimes determined from fits at this wavelength are shown in
Table 2 for 1MA and 3MA. The decay times are equal within
experimental uncertainty to each other and to the decay time
of 9MA. The decay time for 7MA is significantly longer (τ )
4.23 ( 0.13 ps).
Solvent Effects. Pump-probe experiments were carried out
on adenine, 7MA, and 9MA in acetonitrile. 1MA and 3MA
could not be studied due to their low solubility in this solvent,
as mentioned earlier. The transient absorption at 570 nm from
a saturated solution of adenine in acetonitrile (triangles, Figure
6) decays biexponentially as observed for water. Monoexponential decays are seen in Figure 6 for 7MA (circles) and 9MA
(squares). Raw scans are shown in Figure 6 since there is no
significant solvated electron population due to negligible twophoton ionization of acetonitrile. Instead, only a weak residual
offset was observed in this solvent. As in the case of water, the
9MA dynamics are much faster. Decay times were determined
by global fitting to transients recorded at 570, 600, and 630
nm. Table 1 lists values for adenine, while Table 2 contains
lifetimes for 7MA and 9MA in acetonitrile. The trends are the
same as in water, but the lifetimes are somewhat longer,
particularly for 9MA and the fast component seen for adenine.
Finally, experiments were carried out on adenine in unbuffered D2O. The absence of a pH buffer should be of no
(18) Fujii, T.; Itaya, T. Heterocycles 1999, 51, 2255.
Figure 7. Normalized transient absorption signals of adenine in H2O (solid
line) and D2O (O) pumped at 263 nm and probed at 570 nm. The inset
shows the same data at short times.
consequence due to the low pKa of the adenine cation. We
confirmed in a separate experiment that transients for adenine
in unbuffered H2O solution agree completely with transients
obtained in the presence of the phosphate buffer. The transient
signal at 570 nm in D2O is compared with the signal observed
in H2O in Figure 7.
Discussion
Adenine Tautomerism. Adenine has a total of eight structural isomers: four amino isomers in which hydrogen is attached
to any one of the four ring nitrogens and four imino ones. These
are shown in Chart 1 together with their abbreviations. Each of
the four imino tautomers has an E and a Z stereoisomer,
depending whether the hydrogen at the end of the imino CdN
bond is closer to N7 or N1, respectively. There are thus 12
energetically distinguishable isomers of adenine. Following
literature precedent, we will refer to these 12 isomers as
tautomers, although tautomers are, strictly speaking, defined to
be readily interconvertible isomers.19 As will be discussed
below, the energies of many of the possible isomers are
sufficiently high that they do not occur to any significant extent.
The tautomer distribution of adenine in aqueous solution at room
temperature has been studied extensively. There is broad
(19) McNaught, A. D.; Wilkinson, A. Compendium of Chemical Terminology,
2nd ed.; Blackwell Science: Oxford, 1997.
J. AM. CHEM. SOC.
9
VOL. 125, NO. 44, 2003 13597
ARTICLES
Cohen et al.
Chart 2. Amino and Imino Tautomers of n-methyladenine (n ) 1,
3, 7, 9)
consensus from theoretical20-22 and experimental23-26 investigations that two tautomers coexist at significant concentrations
in water: the 7H-amino and 9H-amino forms.
Tautomerism in Monomethylated Adenines. Methylation
at a ring nitrogen reduces the number of tautomers to five: One
amino and two imino forms, each with E and Z conformers.
All possible structures are shown in Chart 2. Tautomerism at
the 7,9 positions of the imidazole ring is blocked for 7MA and
9MA, but can still take place for the imino forms of 1MA and
3MA. As tautomerism has not been studied as extensively for
these compounds as for adenine, we review the experimental
and theoretical evidence for the lowest-energy tautomers in some
detail.
1MA. IR spectra of 1MA in solid Ar indicate that the 9Himino form is the predominant tautomer under matrix isolation
conditions.27 Only very small amounts of the 7H-imino and
amino tautomers could be detected in a later matrix isolation
study.28 Ab initio calculations gave the energetic ordering 9Himino < 7H-imino < amino.27,28 The E conformer of 9H-imino
1MA is lower in energy by 32 kJ mol-1 than the Z conformer.
The latter structure is destabilized by steric repulsion between
the imino hydrogen and the methyl group. On the other hand,
the very similar energies of the E and Z conformers of 1H,7H(20) Holmén, A.; Broo, A. Int. J. Quantum Chem. 1995, 22, 113.
(21) Gu, J.; Leszczynski, J. J. Phys. Chem. A 1999, 103, 2744.
(22) Laxer, A.; Major, D. T.; Gottlieb, H. E.; Fischer, B. J. Org. Chem. 2001,
66, 5463.
(23) Chenon, M.-T.; Pugmire, R. J.; Grant, D. M.; Panzica, R. P.; Townsend,
L. B. J. Am. Chem. Soc. 1975, 97, 4636.
(24) Dreyfus, M.; Dodin, G.; Bensaude, O.; Dubois, J. E. J. Am. Chem. Soc.
1975, 97, 2369.
(25) Gonnella, N. C.; Nakanishi, H.; Holtwick, J. B.; Horowitz, D. S.; Kanamori,
K.; Leonard, N. J.; Roberts, J. D. J. Am. Chem. Soc. 1983, 105, 2050.
(26) Holmén, A. J. Phys. Chem. A 1997, 101, 4361.
(27) Pivovarov, V. B.; Stepanian, S. G.; Reva, I. D.; Sheina, G. G.; Blagoi, Y.
P. Spectrochim. Acta, Part A 1995, 51A, 843.
(28) Schoone, K.; Houben, L.; Smets, J.; Adamowicz, L.; Maes, G. Spectrochim.
Acta, Part A 1996, 52A, 383.
13598 J. AM. CHEM. SOC.
9
VOL. 125, NO. 44, 2003
imino adenine reported by Ha et al.29 suggest that the two 7Himino conformers of 1MA should have similar energies. A
semiempirical study of 1MA tautomerism supports this conclusion.30
The energetic ordering of the 1MA tautomers changes as the
polarity of the environment increases. A UV and IR study by
Dreyfus et al. showed that the amount of the amino form
increases with increasing solvent polarity.31 Thus, the amino
form of 1-heptyladenine increases from ∼20% in chloroform
to ∼50% in methylene chloride.31 In aqueous solution, the
authors estimate that the amino form predominates with only
about 1% imino form present. Ab initio calculations support
the switch from a minimum energy imino to an amino form.
Both the magnitude of the dipole moments (µ ) 3.81 D for the
(E)-9H-imino tautomer of 1 MA vs µ ) 9.59 D for 1MAamino)28 and the comparatively small energy splitting between
the amino and the minimum energy imino form in vacuo (∆E
) 21.1 kJ mol-1)28 are consistent with a lower-energy amino
form in aqueous solution.
3MA. Sharma and Lee recently studied all five tautomers of
3MA at the B3LYP/6-31+G* level of theory.32 For the isolated
compounds, they found that energies increase in the order amino
< (Z)-7H-imino < (E)-7H-imino < (E)-9H-imino ≈ (Z)-9Himino. The energy difference between the amino form and the
next most-stable tautomer of 3MA, the (Z)-7H-imino form, is
43.5 kJ mol-1. The E conformer of the 7H-imino tautomer lies
28.0 kJ mol-1 above the Z conformer. The E conformer is
significantly higher in energy due to steric repulsion between
the imino and the N7 hydrogens. Similar results were reported
in an earlier ab initio study.33 It is interesting to compare the
3MA tautomers with the five analogous tautomers of adenine,
which have a hydrogen atom at N3 in place of the methyl group
in 3MA. From the results of Ha et al.,29 this subset of adenine
tautomers has very similar relative energies as the corresponding
3MA tautomers. This indicates that replacement of N-H by
N-CH3 has a minor effect on the ground-state energies.
Experimental measurements confirm that the amino form of
3MA has the lowest energy in vacuo and in a nonpolar
environment. Thus, the measured gas-phase acidity of 3MA
supports the conclusion that the amino tautomer has the lowest
energy in the gas phase.32 FT-IR spectroscopy revealed the
overwhelming presence of amino 3MA in an argon matrix, with
only a trace amount (∼0.5%) of the 7H-imino tautomer.33
There are several experimental studies on 3MA tautomerism
in solution. Bergmann et al. measured the dipole moment of
3MA to be 4.98 ( 0.10 D in dioxane.34 Since this value was
high compared with a calculated value for the 3H-amino
tautomer of adenine (µ ) 4.2 D35), they speculated that some
amount of the 9H-imino form of 3MA is present due to its
significantly higher dipole moment. However, more recent
calculations have shown that the 9H-imino tautomer lies much
higher in energy above the amino form of 3MA than was found
(29) Ha, T. K.; Keller, M. J.; Gunde, R.; Gunthard, H. H. THEOCHEM 1996,
364, 161.
(30) Kryachko, E. S.; Zundel, G. J. Mol. Struct. 1998, 446, 41.
(31) Dreyfus, M.; Dodin, G.; Bensaude, O.; Dubois, J. E. J. Am. Chem. Soc.
1977, 99, 7027.
(32) Sharma, S.; Lee, J. K. J. Org. Chem. 2002, 67, 8360.
(33) Houben, L.; Schoone, K.; Smets, J.; Adamowicz, L.; Maes, G. J. Mol.
Struct. 1997, 410-411, 397.
(34) Bergmann, E. D.; Weiler-Feilchenfeld, H.; Neiman, Z. J. Chem. Soc. B
1970, 1334.
(35) Pullman, B.; Berthod, H.; Dreyfus, M. Theor. Chim. Acta 1969, 15, 265.
Ultrafast Excited-State Dynamics of Adenines
ARTICLES
to be the case in 1969 by Pullman et al.35 This makes it highly
unlikely that the 9H-imino tautomer could be present in any
significant amount. Additionally, a recent estimate of the dipole
moment of the amino tautomer of 3MA (µ ) 4.74 D33) shows
that there is no longer a significant discrepancy with the
experimental value of Bergmann. In a separate study in solution,
the lack of an anionic pKa for 3-ethyladenine in water was
suggested by Pal to favor the amino tautomer.36 However,
Dreyfus et al. have pointed out the dangers of using pKa data
to resolve questions about tautomerism.31 Dreyfus et al. did not
observe an imidazole N-H stretch in IR spectra of 3-methyladenine and 3-benzyladenine in chloroform.31 This result and
the absence of a tautomeric equilibrium in temperature-jump
experiments in aqueous solution led them to conclude that 3MA
exists in the amino form regardless of enVironment.
To our knowledge, there are no theoretical studies of 3MA
tautomerism that include solvation. However, studies on adenine
have shown that the size of the molecular dipole moment has
the largest effect on the solvation energy.20,22 The dipole moment
of (Z)-7H-imino (µ ) 3.70 D) is somewhat smaller than the
value for the amino form (µ ) 4.74 D), according to the
calculations of Houben et al.33 The (E)-7H-imino tautomer has
a larger dipole moment of 5.61 D,33 but lies 70-80 kJ mol-1
above the amino form. Thus, solvation is not expected to
significantly stabilize either of the next lowest-energy tautomers
of 3MA. We conclude that available experimental and theoretical evidence indicates that 3MA is present overwhelmingly as
the amino tautomer in aqueous solution.
7MA and 9MA. To our knowledge, there are no reports of
imino tautomer energetics for 7MA or 9MA. However, replacing
the methyl group of each compound by hydrogen generates an
isomer of adenine, which has been studied. As shown above,
calculated energies for 1MA and 3MA tautomers are very close
to corresponding adenine ones. The results from Ha et al.29 on
all 12 adenine tautomers can therefore be used to predict
energetics for the various tautomers of 7MA and 9MA. For
example, the 1H,9H-imino tautomer of adenine (E and Z
conformers have approximately the same energy) lies 51.3 kJ
mol-1 above the 9H-amino tautomer, while the 3H,9H-imino
tautomer lies 136 kJ mol-1 higher in energy.29 The 1H,7H-imino
tautomer lies 36.9 kJ mol-1 above the 7H-amino tautomer of
adenine.29 These values, coupled with the modest dipole
moments of the lowest energy tautomers, predict that the 7MA
and 9MA will be present exclusively in the amino form.
Experiments strongly support this conclusion.34,37,38
In summary, the evidence indicates that 1MA, 3MA, 7MA,
and 9MA are each present in a single tautomeric form in
aqueous solution, the amino form. In contrast, adenine is present
as two amino tautomers (referred to below as the 7H and 9H
tautomers) under the same conditions.
Biexponential Dynamics of Adenine. The signal for adenine
in H2O shows a striking, biexponential decay. In contrast,
transient absorption by the adenine nucleoside, adenosine, shows
only a monoexponential decay with a lifetime of 290 fs.2
Biexponential decay was also observed for adenine in fluorescence up-conversion experiments by Gustavsson et al., who
reported lifetimes of 0.23 ( 0.05 ps and 8.0 ( 0.3 ps at a
temperature of 20 ( 1 °C.4 These values are in excellent
agreement with the lifetimes in Table 1, demonstrating that our
transient absorption experiments monitor the decay of the
fluorescent state. Clearly, these results refute an early timeresolved report in which it was claimed that there is little change
to the fluorescence lifetime of adenine upon ribosyl substitution.39 From global fits at several wavelengths, decay times of
4.23 ( 0.13 ps and 0.22 ( 0.02 ps were determined for 7MA
and 9MA in H2O, respectively. These values clearly resemble
the long (τ ) 8.8 ( 1.2 ps) and short (τ ) 0.18 ( 0.03 ps)
decay components seen for adenine. Since adenine is only
expected to be present as the 7H and 9H tautomers in aqueous
solution, we attribute the long and short decays to these species,
respectively. The lifetimes for the 9H tautomer and 9MA are
equal within experimental uncertainty, indicating that the
electronic structure of these two species must be very similar.
Recent molecular beam experiments have revealed similar
REMPI spectra for 9-methyladenine and 9H-amino adenine,40
suggesting that the monomethylated compounds are good
models for the electronic properties of the corresponding
H-substituted tautomers of adenine. In contrast, changing methyl
to hydrogen at the N7 position of adenine increases the lifetime
by a factor of 2. This effect is poorly understood at present, but
is indicative of a special sensitivity to N7 substitution, as
discussed more fully below.
Quantitative Estimate of the Adenine Tautomer Distribution in H2O. The two decay components seen for adenine in
water are due to the presence of the 7H and 9H tautomers. In
this case, the decay amplitudes can be used to estimate the
amount of each tautomer. It is necessary, however, to account
for differences in the ground- and excited-state cross sections
for absorption. The signal from a mixture of two tautomers
depends on their relative abundance, as well as on their
respective ground- and excited-state absorption cross sections.
The transient absorption signal, ∆A(t), for adenine can be
modeled as
(36) Pal, B. C. Biochemistry 1962, 1, 558.
(37) Kistenmacher, T. J.; Rossi, M. Acta Crystallogr., Sect. B 1977, B33, 253.
(38) McMullan, R. K.; Benci, P.; Craven, B. M. Acta Crystallogr., Sect. B 1980,
B36, 1424.
(39) Nikogosyan, D. N.; Angelov, D.; Soep, B.; Lindqvist, L. Chem. Phys. Lett.
1996, 252, 322.
(40) Lührs, D. C.; Viallon, J.; Fischer, I. Phys. Chem. Chem. Phys. 2001, 3,
1827.
∆A(t) ∝ f77˜ 7 exp(-t/τ7) + f99˜ 9 exp(-t/τ9)
(1)
In eq 1, the subscripts 7 and 9 label the 7H and 9H tautomers
of adenine, f is the fractional population, τ is the lifetime of
each species, is the molar absorption coefficient in the
electronic ground state at the pump wavelength, and ˜ is the
molar absorption coefficient in the S1 state at the probe
wavelength. Fitting the adenine signal in H2O to two exponentials gives the prefactors for both exponentials in eq 1. The
amplitude associated with the slower decay (A2 in Table 1) is
proportional to f77˜ 7, while the amplitude of the subpicosecond
decay (A1 in Table 1) is proportional to f99˜ 9. Thus, the
fractional population of the 7H tautomer is given by
(
f7 ) 1 +
)
A17˜ 7
A29˜ 9
-1
(2)
According to Dreyfus et al., extinction coefficients of the 7H
and 9H tautomers of adenine are given approximately by shifting
the corresponding values of 7MA and 9MA by 2.5 nm to shorter
J. AM. CHEM. SOC.
9
VOL. 125, NO. 44, 2003 13599
ARTICLES
Cohen et al.
wavelengths.24 The wavelength-dependent extinction coefficients
for 7MA and 9MA are shown in the inset to Figure 1. Following
this procedure, we estimate that 7/9 is equal to 0.85 at our
pump wavelength of 263 nm. The ratio of the ESA cross
sections, ˜ 7/˜ 9, was estimated at a probe wavelength of 570 nm
from the back-to-back scans on 7MA and 9MA shown in Figure
4. Since the solutions had the same absorbance at the pump
wavelength, the ratio of the signal amplitudes is equal to the
ratio of the ESA cross sections. Although the signal from 9MA
appears slightly weaker than the one from 7MA in Figure 4,
this is an artifact of the finite time resolution. Signals with time
constants that are comparable to or shorter than the instrumental
response time are strongly attenuated. The correct amplitudes
are obtained from the nonlinear least-squares fit, which was
carried out with convolution. The deconvoluted amplitude of
the 9MA signal is ∼50% larger than the one from 7MA, and
the ratio ˜ 7MA/˜ 9MA is estimated to be 0.66. We have assumed
that ˜ 7MA/˜ 9MA ≈ ˜ 7/˜ 9. This assumption seems reasonable in
view of the negligible change to the ground-state spectra upon
replacing a hydrogen atom with a methyl group.24 Inserting these
values and the amplitudes A1 and A2 from Table 1 into eq 2
results in 22% for the fractional population of the 7H tautomer.
Using a conservative estimate of 20% for the fractional error
in ˜ 7/˜ 9, we calculate a standard deviation of 0.036 for f7. The
quantity ˜ 7/˜ 9 is subject to the greatest uncertainty and dominates
the error analysis.
Our best estimate for the population of the 7H tautomer (22
( 4%, 1σ) can be compared with several previous measurements. The distribution of adenine tautomers was studied by
carbon-13 NMR spectroscopy by Chenon et al.23 Prototropic
tautomerism in the imidazole ring of adenine was inferred from
the chemical shifts of the C4 and C5 carbon atoms. The authors
determined the populations of the 7H and 9H tautomers to be
15 and 85%, respectively. Using nitrogen-15 NMR spectroscopy, Gonnella et al. likewise determined from chemical shifts
that aqueous solutions are 20% 7H-amino and 80% 9H-amino.25
In a separate study, Dreyfus et al. determined a value of 22%
for the 7H via temperature-jump and UV absorption experiments.24 The 7H tautomer was determined to be present at a
concentration of 23 ( 3% in poly(vinyl alcohol) film at room
temperature by IR spectroscopy.26
Solvent Effects and Mechanistic Implications. Table 2
shows that decay by 9MA is faster in H2O than in acetonitrile,
while the 7MA lifetime is more similar in the two solvents.
Similar trends are found when comparing the fast and slow
components of the adenine decays in both solvents. With the
exception of 7MA, going from water to acetonitrile increases
the lifetime of each compound by 50-100%. The ratio A1/A2
(see Table 1) is slightly larger in acetonitrile than in water. This
suggests that the percentage of the 7H tautomer may decrease
slightly in acetonitrile. However, a more precise statement is
not possible since we were unable to measure relative ESA cross
sections in acetonitrile since the solubility of adenines in this
solvent prevented us from preparing equal absorbance solutions.
Eastman and Rosa wrote that the percent 7H tautomer should
increase with increasing strength of solvent hydrogen bonds.41
The lack of a dramatic change in acetonitrile suggests this view
may be incorrect. Ab initio studies of solvent effects have shown
that inclusion of specific water molecules does not significantly
(41) Eastman, J. W.; Rosa, E. J. Photochem. Photobiol. 1968, 7, 189.
13600 J. AM. CHEM. SOC.
9
VOL. 125, NO. 44, 2003
change the tautomer energies.21 Gonnella et al. argued that the
7H-amino tautomer of adenine is more strongly favored in water
than in dimethyl sulfoxide.25 Using alkylated adenines to model
the tautomeric equilibrium in adenine, they estimated that the
percentage of the 7H-amino decreases from 20% in water to
14% in DMSO. In our view, specific solute-solvent interactions
such as hydrogen bonding are less important than continuum
solvation of the highly polar 7H tautomer by the bulk liquid.
Changing the solvent from H2O to D2O produces no change
in the transient signals within our experimental uncertainty, as
shown in Figure 7. It is not surprising that the amplitudes of
the slow and fast components are the same in both solvents
since tautomer energetics are independent of deuteration,
neglecting zero-point energies. On the other hand, the absence
of a solvent kinetic isotope effect has profound implications
for the nonradiative decay mechanism. Phototautomerization has
been suggested as a mechanism for ultrafast excited-state decay
by isocytosine,42 cytosine,43 uracil,44 and adenine.45,46 An
excited-state proton-transfer reaction would presumably show
a measurable isotope effect. The absence of such an effect argues
strongly against phototautomerization. In addition, if solventassisted hydrogen transfer were important, then a larger solvent
effect should be observed upon going from water to the aprotic
solvent, acetonitrile. Finally, if a solvent-assisted hydrogen shift
were the primary nonradiative decay mechanism, then S1
lifetimes ought to be considerably longer for isolated nucleobases due to greater energetic barriers. In fact, recent evidence
suggests that ultrashort lifetimes can be observed for single bases
in the complete absence of a solvent.15,40,47
Salter and Chaban recently studied minimum energy reaction
paths for excited-state hydrogen atom transfer from N9 to N3.46
The barrier height for this reaction was found to be lower and
the shape of the reaction path to be more gently sloping for the
excited state. This suggests that tautomerization on the excitedstate potential energy surface is more probable than on the
ground state. The authors pointed out that one or more water
molecules could dramatically lower the barrier, as observed in
other proton-transfer reactions. Our observation of ultrafast
lifetimes (τ ≈ 200 fs) for 3MA and 9MA, neither of which can
undergo a 9 f 3 hydrogen shift, makes this mechanism unlikely
to be the dominant nonradiative decay pathway.
Domcke and Sobolewski recently proposed an intriguing
mechanism for ultrafast nonradiative decay in adenine, which
they suggest applies in vacuo and in solution.11,48 Specifically,
they identified a 1πσ* state by ab initio calculations, which
electronically predissociates the 1ππ* and 1nπ* states along an
N-H stretching coordinate. At a large value of the N-H
coordinate, the 1πσ* state has a conical intersection with the
electronic ground state, closing a pathway for ultrafast internal
conversion. The authors presented reaction-path potential energy
profiles for this mechanism as a function of the N9-H distance
in 9H-adenine.11,48 However, the present results on 9MA and
our previous measurements on adenosine1,2 clearly indicate that
subpicosecond relaxation is possible in the absence of this
(42)
(43)
(44)
(45)
Shukla, M. K.; Leszczynski, J. Int. J. Quantum Chem. 2000, 77, 240.
Shukla, M. K.; Leszczynski, J. J. Phys. Chem. A 2002, 106, 11338.
Shukla, M. K.; Leszczynski, J. J. Phys. Chem. A 2002, 106, 8642.
Mishra, S. K.; Shukla, M. K.; Mishra, P. C. Spectrochim. Acta, Part A
2000, 56, 1355.
(46) Salter, L. M.; Chaban, G. M. J. Phys. Chem. A 2002, 106, 4251.
(47) Kang, H.; Jung, B.; Kim, S. K. J. Chem. Phys. 2003, 118, 6717.
(48) Sobolewski, A. L.; Domcke, W. Eur. Phys. J. D 2002, 20, 369.
Ultrafast Excited-State Dynamics of Adenines
coordinate. Adenosine, 1MA, 3MA, and 9MA all have subpicosecond lifetimes despite lacking an N-H bond on the purine
ring. It is still conceivable that the Domcke-Sobolewski mechanism could instead involve one of the N-H bonds on the
exocyclic amino group, but the energies of the 1πσ* state
associated with this group have not been determined. Moreover,
it is difficult to envision how a 1πσ* state associated with the
amino group could interact strongly with 1ππ* and 1nπ* states
localized on the ring(s).
We believe the most probable explanation for the femtosecond
lifetimes of the DNA bases is the proximity of the 1nπ* and
1ππ* states.49 Several authors have discussed the effect of
proximal 1ππ* and 1nπ* states on the photophysics of adenine7,8,10,50 and 2-aminopurine.9 For adenine, these studies make
a plausible case for a state switch from the 1ππ* surface to a
1nπ* state, but they do not explain ultrafast nonradiative decay
to the electronic ground state. By detecting hot ground state
absorption we have shown previously that S0 is repopulated on
a subpicosecond time scale.1,2 If the proximity mechanism is
correct, then it must also explain ultrafast relaxation from the
1nπ* state to the electronic ground state.48 The recent computational study of Ismail et al. located conical intersections
between the two 1nπ* states of cytosine and the electronic
ground state.12 This mechanism appears sufficiently universal
to account for ultrafast decay by the bases studied here and in
ref 16. It is also consistent with the solvent effects reported
here. Thus, changing the solvent polarity can differentially affect
the energies of 1ππ* and 1nπ* states,9,10 leading to changes in
coupling. We believe that this polarity change and not the
change in hydrogen bonding underlies the modest difference
in lifetimes seen in water and in acetonitrile. Finally, the absence
of a solvent kinetic isotope effect is not inconsistent with decay
by conical intersections. Such features appear to be accessed
by motion along complex multidimensional coordinates.12
A great deal of computational work is required to confirm
the mechanistic ideas discussed here. Our hope is that this study
can provide significant tests of emerging theoretical models for
nonradiative decay. Interestingly, the attachment of -H, -methyl,
or -ribose at N9 has minimal effect on the lifetimes. From the
pattern of decay by the monomethylated adenines, it appears
that substitution at N7 is the only factor correlated with longer
fluorescence lifetimes. This is consistent with the dramatically
increased fluorescence lifetime of the guanosine cation,2 which
is protonated at N7. Ismail et al. showed that the two 1nπ*
excitations for cytosine do not lie at the same energy.12 Our
results suggest that the 1nπ* state associated with the N7 lone
pair plays a decisive role in the singlet excited-state deactivation.
Removal of this state in the 7-substituted compounds may extend
the lifetime due to weaker coupling with a different 1nπ* state,
which lies higher in energy. In support of this picture, we cite
recent TDDFT calculations which have shown that the minimumenergy 1nπ* state lies higher above the lowest-energy 1ππ* state
in 7H than in 9H adenine.10
Kim and co-workers recently studied adenine and several
adenine derivatives under isolated molecule conditions by
ARTICLES
femtosecond pump-probe transient ionization time-of-flight
mass spectrometry.47 They reported lifetimes of 1.00 ( 0.05
ps, 1.06 ( 0.08 ps, and 0.94 ( 0.03 ps for adenine, 7MA, and
9MA, respectively, upon excitation with femtosecond pulses at
267 nm. Their signals were monoexponential in all cases. The
monoexponential signal seen by these workers for adenine
contrasts with the biexponential signal reported here in aqueous
solution, but is not unexpected since the 9H-amino tautomer of
adenine is the only form present in the gas phase.40 The
monoexponential decays Kang et al. observed for 7MA and
9MA are consistent with the monoexponential decays we
observe in solution. The short lifetimes seen even for isolated
molecules suggest that the solvent does not play a significant
role in excited-state decay. The finding by Kang et al.47 of
identical lifetimes for 7MA and 9MA is surprising in light of
our lifetimes of 4.23 and 0.22 ps for these compounds,
respectively, in room temperature aqueous solution. Since
solvent interactions generally quench excited states, it is difficult
to explain the decreased lifetime of 7MA in the molecular beam,
and further work is needed before the solution-phase measurements can be reconciled with those in the gas phase. It is worth
noting that excited-state lifetimes for isolated nucleobases appear
to depend sensitively on vibrational excess energy in the excited
state. Thus, considerably longer lifetimes have been observed
in gas-phase experiments on DNA bases for excitation much
closer to the electronic origin.40,51
Conclusions
The femtosecond transient absorption technique has been used
to study the solvent- and tautomer-dependent photophysics of
adenine. The room-temperature signals from adenine in aqueous
solution are biexponential due to an equilibrium mixture of 7H(∼22%) and 9H-amino (∼78%) tautomers. The S1 lifetime of
7H-adenine is approximately 40 times longer than that of the
more abundant 9H-tautomer, and this is the reason the timeintegrated emission arises principally from the former isomer.
This study shows definitively that individual tautomers of the
same nucleobase can differ greatly in their excited-state dynamics. Chin et al. have reached the same conclusion from gasphase experiments on guanine.51 The results for adenine and
for the monomethylated adenines demonstrate that substitution
at N7 in the imidazole ring has a more pronounced effect on
the excited-state decay than substitution at any of the other ring
nitrogens. We have shown previously that the fluorescence
lifetimes of cytosine derivatives are exquisitely sensitive to
chemical substitution.16 The work presented here extends this
sensitivity to individual prototropic tautomers. In view of the
sensitivity of nucleobase excited-state dynamics to covalent
modification, an open question for the future is the extent to
which the noncovalent interactions (i.e., base stacking and base
pairing) found in nucleic acid polymers affect their excitedstate dynamics.
Acknowledgment. This publication was made possible by
NIH Grant No. 1 R01 GM64563-01 from the National Institute
of General Medical Sciences.
JA035628Z
(49) Lai, T. I.; Lim, B. T.; Lim, E. C. J. Am. Chem. Soc. 1982, 104, 7631.
(50) Holmén, A.; Broo, A.; Albinsson, B.; Nordén, B. J. Am. Chem. Soc. 1997,
119, 12240.
(51) Chin, W.; Mons, M.; Dimicoli, I.; Piuzzi, F.; Tardivel, B.; Elhanine, M.
Eur. Phys. J. D 2002, 20, 347.
J. AM. CHEM. SOC.
9
VOL. 125, NO. 44, 2003 13601